The Performance Counter Monitor can watch any performance counter that the Windows Perfmon tool can display. This gives you great flexibility since many systems and drivers on the computer report statistics and their current state via the performance counters.
In addition, CPU and memory usage counters are simulated for Linux/Unix machines via SNMP, and via VMWare interfaces for ESX servers. Monitoring and charting those values is now as easy as with a Windows server.
When adding a counter to be monitored, the target server is queried for counters it supports, and a dialog similar to the Windows Perfmon dialog is shown where you can select the counter that you want to monitor. Servers monitored by Satellites will also be queried immediately for their counter lists. For Linux/Unix servers, SNMP is used. Make sure you have entered the SNMP credentials if the defaults of v2c and 'public' won't work.
Once a counter has been chosen, alerting criteria are specified. A threshold (low or high) and the amount of time the threshold has to be exceeded before actions are fired can be given. If the theshold is passed, the offending counter and its current value will be part of the action description.
Each time a performance counter is measured and checked, it is also recorded in a database in order to generate historical reports.
The Performance monitor can create charts from any of the counter values that are being monitored. This includes bar and line charts. Tabular HTML and .CSV output for importing into other apps (Excel) are also possible.
Windows doesn't have a performance counter for % Network Utilization, so PA Server Monitor has support for computed counters for Send and Receive Utilization percents. To get them, you need to add the two base counters that each is based on. The computed values are percentages from 0 to 100.
For each of the counters below, you can set the alert thresholds to anything you want. Current Bandwidth is supposed to be a constant value, so monitoring it for alerting purposes isn't important. In that case an alert threshold of = 0 would be a good default. For Bytes Sent and Received/sec you also might not care about alerting on it. Since it could realistically be 0 some times, you could set the alert threshold to < 0 so it never alerts.
To get \Network Interface\(<network card>)\% Receive Utilization [Ex: alert on > 95]
To get \Network Interface\(<network card>)\% Send Utilization [Ex: alert on > 95]